Twin studies provide estimates of genetic and environmental contributions to cognitive ability differences, but could be based on biased samples. Here we report whole-population estimates using twins from unique mental surveys in Scotland. The Scottish Mental Surveys of 1st June 1932 (SMS1932) and 4th June 1947 (SMS1947), respectively, administered the same validated verbal reasoning test to almost everyone born in 1921 or 1936 and attending school in Scotland. There were 572 twin pairs from the SMS1932, and 517 pairs from the SMS1947. Information on zygosity was unavailable. A novel application of a mixture distribution was used to estimate genetic and environmental components of verbal reasoning variation by maximum likelihood. We found consistent heritability (approximately 0.70) and shared environment (approximately 0.21) estimates. The estimates did not change substantially when additional quantitative traits (height and weight) were added in a multivariate analysis. More generally for studies in genetics, the methodological innovation developed here implies that large (national) data collections can provide sufficient information on twin pairs to estimate genetic parameters, even without zygosity.